Robust Quantum Memory in a Trapped-Ion Quantum Network Node P. DrmotaD. Main D. P. Nadlinger B. C. Nichol M. A. Weber E. M. Ainley A. Agrawal R. Srinivas G. Araneda C. J. Ballance and D. M. Lucas

2025-05-03 0 0 912.94KB 10 页 10玖币

侵权投诉

Robust Quantum Memory in a Trapped-Ion Quantum Network Node

P. Drmota,∗D. Main, D. P. Nadlinger, B. C. Nichol, M. A. Weber, E. M. Ainley,

A. Agrawal, R. Srinivas, G. Araneda, C. J. Ballance, and D. M. Lucas

Department of Physics, University of Oxford, Clarendon Laboratory,

Parks Road, Oxford OX1 3PU, United Kingdom

We integrate a long-lived memory qubit into a mixed-species trapped-ion quantum network node.

Ion-photon entanglement ﬁrst generated with a network qubit in 88Sr+is transferred to 43 Ca+with

0.977(7) ﬁdelity, and mapped to a robust memory qubit. We then entangle the network qubit with

a second photon, without aﬀecting the memory qubit. We perform quantum state tomography to

show that the ﬁdelity of ion-photon entanglement decays ∼70 times slower on the memory qubit.

Dynamical decoupling further extends the storage duration; we measure an ion-photon entanglement

ﬁdelity of 0.81(4) after 10 s.

Quantum networks have the potential to revolutionize

the way we distribute and process information [1]. They

have applications in cryptography [2, 3], quantum com-

puting [4, 5], and metrology [6], and can provide insights

into the nature of entanglement [7, 8]. Photonic inter-

faces are essential for such networks, enabling two remote

stationary qubits to exchange quantum information using

entanglement swapping [9]. Elementary quantum net-

works have been realized with diamond nitrogen-vacancy

centers [8, 10], photons [11, 12], neutral atoms [13–15],

solid-state systems [16], and trapped ions [7, 17–24].

Trapped ions provide qubits with exceptionally long

coherence times, which can be initialized, manipulated,

entangled, and read out with high ﬁdelity [25–30]. More-

over, trapped ions readily interact with optical ﬁelds,

providing a natural interface between their electronic

state – the stationary quantum memory – and photons

– the “ﬂying” quantum information carrier [31]. Pairs of

trapped-ion network nodes comprising one qubit of a sin-

gle species have been connected by a photonic link and

used to perform Bell tests [7], state teleportation [18],

random number generation [19], quantum key distribu-

tion [21], and frequency comparisons [22]. Trapped ion

systems have also demonstrated state-of-the-art single-

and two-qubit gate ﬁdelities, but integrating these within

a quantum network node remains a challenge since an

ion species suitable for quantum communication does not

necessarily also provide a good memory qubit with suf-

ﬁcient isolation from network activity. Atomic species

such as 133Ba+or 171Yb+have been proposed to cir-

cumvent this issue [26, 32]; however, the development of

the required experimental techniques is still ongoing. Al-

ternatively, it is possible for each role to be fulﬁlled by a

diﬀerent species [33]. In addition, using multiple atomic

species has advantages for minimizing crosstalk during

mid-circuit measurements and cooling [34].

In this Letter, we demonstrate a trapped-ion quan-

tum network node in which entanglement between a net-

work qubit and a photon is created and coherently trans-

ferred onto a memory qubit for storage, while the net-

work qubit is entangled with a second photon. Due

to its simple level structure, 88Sr+is ideally suited for

our ion-photon entanglement (IPE) scheme [20], whereas

the hyperﬁne structure of 43Ca+provides a long-lived

memory qubit [35]. While both IPE and local mixed-

species entangling gates have been demonstrated inde-

pendently [33], this is the ﬁrst experiment in which these

capabilities are combined. Furthermore, we show that

the memory qubit in 43Ca+is robust to environmental

noise as well as to concurrent addressing of 88Sr+for the

generation of IPE. Finally, sympathetic cooling of the ion

pair using 88Sr+between rounds of entanglement gener-

ation enables continued operation even in the presence of

heating.

For this experiment, we load a 88Sr+-43Ca+crystal

with controlled order into a surface-electrode Paul trap

at room temperature [36]. Each experimental sequence

begins with cooling [37], reducing the temperature of the

axial out-of-phase (OOP) and in-phase (IP) motion to

¯noop '0.3 and ¯nip '3, respectively. The cooling se-

quence was empirically optimized for the high heating

rates observed, namely ˙

¯noop '360 s−1at ωoop/(2π) =

3.354 MHz and ˙

¯nip '2700 s−1at ωip/(2π)=1.705 MHz.

To produce single photons, 88Sr+is excited to the

|P1/2, mJ=+1/2istate by a ∼10 ps laser pulse. This

short-lived excited state decays with probability 0.95 into

the S1/2manifold via emission of a photon at 422 nm

whose polarization is entangled with the spin state of the

ion. The photon emission is imaged by an NA = 0.6 ob-

jective onto a single-mode optical ﬁber [Fig. 1(a)], which

acts as a spatial mode ﬁlter, maximizing the entangled

fraction by suppressing polarization mixing. The ion-

photon state can then be described by the maximally

entangled Bell state

|ψi=1

√2|↓Ni⊗|Hi+|↑Ni⊗|Vi,

where |Hiand |Viare orthogonal linear polarization

states of the photon, and |↓Niand |↑Nidenote the net-

work qubit states in the Zeeman ground state manifold

of 88Sr+[Fig. 1(b)]. To analyze the polarization state

of the photon, we employ polarizing beamsplitters and

avalanche photodiodes, which are part of the same pho-

arXiv:2210.11447v2 [quant-ph] 7 Apr 2023

tonic Bell state analyzer used to herald remote entangle-

ment between two network nodes [20]. The pulsed excita-

tion sequence is repeated in a loop at an attempt rate of

1 MHz until a photon is detected. The polarization mea-

surement basis is set at the beginning of a sequence us-

ing motorized waveplates. Qubit manipulation of 88Sr+

674 nm

402 nm

422 nm

402 nm

Bell State

Analyzer

NA=0.6

(a)

S1/2

F= 3

F= 4

|↑M⟩

···

|↑L⟩

|↓M⟩

···

|↓L⟩

3.2 GHz

P1/2

P3/2

43Ca+

|↓N⟩

|↑N⟩

S1/2

14 MHz

P3/2

P1/2

D5/2

674 nm

402 nm

422 nm

88Sr+

(b)

FIG. 1. (a) Overview of the apparatus. We show the laser

beam geometry; within the plane of the trap surface, the mag-

netic ﬁeld Bis oriented at 45°to the trap axis z. Perpendic-

ular to this plane, the NA = 0.6 lens collects single photons

from a 88Sr+ion (violet circle). Single photons are coupled

into a single-mode ﬁber that is connected to a Bell state an-

alyzer. Here, only one network node is connected; the same

device can herald remote entanglement with a second, iden-

tical node [20]. The state of 88Sr+can be mapped onto a

co-trapped 43Ca+ion (orange circle). (b) Level structure of

88Sr+(violet) and 43Ca+(orange), not to scale. The mem-

ory qubit comprises the mF= 0 states in the 43Ca+S1/2

manifold. Raman lasers (blue arrows, 422 nm) are used to

drive mixed-species entangling gates and transitions between

43Ca+hyperﬁne ground states. A σ+-polarized laser pulse

excites the S1/2↔P1/2transition in 88Sr+to generate a sin-

gle photon whose polarization (see σand πdecay channels) is

entangled with the state of the ion. A narrow-linewidth laser

(red arrow, 674 nm) is used to manipulate the 88Sr+qubit via

the quadrupole transition.

is performed on the 674 nm quadrupole transition, which

is also used for electron shelving readout.

The second ion species, 43Ca+, is chosen for its excel-

lent coherence properties and the high level of control

achieved in previous experiments [27, 38–40]. Further-

more, the mass ratio between 43Ca+and 88Sr+is reason-

ably favorable for sympathetic cooling [41], and the elec-

tronic level structure facilitates mixed-species gates [42].

For state preparation, polarized 397 nm light optically

pumps population into |↓Li. A pair of co-propagating

Raman laser beams at λR= 402 nm is used to manipu-

late states within the ground state manifold. For readout,

population is shelved using a pulse sequence of 393 nm

and 850 nm light [43]. At a magnetic ﬁeld of 0.5 mT,

the frequency of the memory qubit transition depends

weakly on the magnetic ﬁeld magnitude, with a sensitiv-

ity of 122 kHz mT−1. Compared to the sensitivity of the

88Sr+Zeeman qubit of 28 MHz mT−1, the memory qubit

is signiﬁcantly more resilient to magnetic ﬁeld noise. In

addition, the magnetic ﬁeld at the position of the ions

is actively stabilized using feedback and feedforward cur-

rents to the 10 nT level, calibrated over the 50 Hz line

cycle using 88Sr+as a magnetic ﬁeld probe [44].

To swap information from 88Sr+to 43Ca+, we per-

form mixed-species ˆσz⊗ˆσzgeometric phase gates us-

ing the state-dependent light shift force generated by a

single pair of ∼20 mW Raman laser beams at 402 nm.

Only one pair is required to drive both species thanks to

their compatible electronic level structure [42] [Fig. 1(b)].

The main advantage of this scheme over Cirac-Zoller and

Mølmer-Sørensen gates, which have previously been ex-

plored in this context [33, 45], is its robustness to qubit

frequency shifts. The Raman beams are aligned to ad-

dress the OOP axial motion of the two-ion crystal [42].

For maximum gate eﬃciency on this mode, the ion spac-

ing is set to a half-integer multiple of the eﬀective wave-

length λR/√2. As the memory qubit in 43Ca+does not

experience a light shift, this interaction is performed on

the logic qubit L instead. First-order Walsh modulation

compensates for the light shift imbalance between the

two species. With the available laser power, a detuning

of δ/(2π) = 34 kHz from the OOP mode achieves a gate

duration of ≈60 µs while minimizing oﬀ-resonant excita-

tion of the IP mode [37]. Charging of the trap due to the

Raman laser beams is automatically compensated every

∼5 min using the method described in Ref. [46].

The state of the network qubit in 88Sr+is coherently

swapped onto the logic qubit using an iSWAP gate, which

is implemented by two ˆσz⊗ˆσzinteractions and single-

qubit rotations [circuit shown in Fig. 2(a)]. We use the

iSWAP, as opposed to a full SWAP, since the initial state

of 43Ca+is known to be prepared in |↓Li. The ideal

iSWAP performs the mapping |φNi⊗|↓Li 7→ |↓Ni⊗|φLi,

where |φiis the arbitrary state to be swapped to the

logic qubit, leaving the network qubit in the |↓Nistate.

Experimental imperfections leading to deviations from

文档加载中……请稍候！
如果长时间未打开，您也可以点击刷新试试。

下载文档到电脑，查找使用更方便

10 玖币 0人已下载

立即下载

摘要：

RobustQuantumMemoryinaTrapped-IonQuantumNetworkNodeP.Drmota,D.Main,D.P.Nadlinger,B.C.Nichol,M.A.Weber,E.M.Ainley,A.Agrawal,R.Srinivas,G.Araneda,C.J.Ballance,andD.M.LucasDepartmentofPhysics,UniversityofOxford,ClarendonLaboratory,ParksRoad,OxfordOX13PU,UnitedKingdomWeintegratealong-livedmemoryqubitin...

展开>> 收起<<

Robust Quantum Memory in a Trapped-Ion Quantum Network Node P. DrmotaD. Main D. P. Nadlinger B. C. Nichol M. A. Weber E. M. Ainley A. Agrawal R. Srinivas G. Araneda C. J. Ballance and D. M. Lucas.pdf

共10页,预览2页

还剩页未读，继续阅读

声明：本站为文档C2C交易模式，即用户上传的文档直接被用户下载，本站只是中间服务平台，本站所有文档下载所得的收益归上传人(含作者)所有。玖贝云文库仅提供信息存储空间，仅对用户上传内容的表现方式做保护处理，对上载内容本身不做任何修改或编辑。若文档所含内容侵犯了您的版权或隐私，请立即通知玖贝云文库，我们立即给予删除！

Robust Quantum Memory in a Trapped-Ion Quantum Network Node P. DrmotaD. Main D. P. Nadlinger B. C. Nichol M. A. Weber E. M. Ainley A. Agrawal R. Srinivas G. Araneda C. J. Ballance and D. M. Lucas

相关推荐

开通VIP享超值会员特权

作者详情

相关内容

热门标签

举报选择: